A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of any portion of the patent document, as it appears in any patent granted from the present application or in the Patent and Trademark Office file or records available to the public, but otherwise reserves all copyright rights whatsoever.
An appendix containing source code listing and related materials utilized in practicing an exemplary embodiment of the invention is included as part of the Specification and is hereinafter referred to as Appendix A.
The present invention relates to a resistance-capacitance (R-C) balancing network for a hybrid for use in a full-duplex telecommunication system, to a method for optimizing transhybrid loss, and to methods and apparatus for synthesizing impedances and simulating telecommunications circuits. The invention is particularly useful in the design, operation, and simulation of echo cancelers used in telecommunication systems.
Full duplex two wire telecommunications systems use a single wire-pair to transmit and receive data. The transmitted and received data often have overlapping spectra or frequency bands. As such, it is necessary to separate transmitted and received signals to facilitate the full duplex data transfer. A technique used to achieve this separation is known as echo cancellation. The echo cancellation function can be performed by a full-duplex to half-duplex conversion circuit, for example, a two-wire (xe2x80x9c2Wxe2x80x9d) to four-wire (xe2x80x9c4Wxe2x80x9d) converter, or hybrid.
Once the transmit and receive directions of transmission have been separated, they can be transmitted over facilities on independent half-duplex paths. The independent half-duplex paths are known as a four-wire facility.
Communications systems can use a combination of digital and analog echo cancelling techniques. One embodiment of the present invention concentrates on analog techniques, but does not preclude the adaptation of those techniques to digital forms as is readily understood by those of skill in the art.
The task of an echo canceller is to cancel the near-end signal sufficiently well so that the far end signal can be detected at the near-end. This task is often especially challenging as the far end signal has undergone significant attenuation in the intervening cable which connects near and far ends.
Performance of the echo cancellation task requires, in some form, knowledge of the impedance seen looking into the cable pair. In accordance with the invention, I describe simple networks which accurately model this impedance, as well as methods for designing those networks.
When these networks are used in hybrids, they are often referred to as balancing networks. When used in other echo cancelling structures, these networks may have other names, but their function is still similar, i.e., to aid the cancellation of near end signal at the near end, while maximizing the amount of far end signal received at the near end.
In systems using a balancing network, the quality of the impedance match between the balancing network and the local loop is measured by the transhybrid loss of the conversion circuit, e.g., a hybrid, which loss is a measure of the effectiveness of the network. Transhybrid loss (xe2x80x9cTHLxe2x80x9d) is typically expressed in decibels (xe2x80x9cdBxe2x80x9d) by equation (1):                     THL        =                  20          ⁢                      xe2x80x83                    ⁢          log          ⁢                      "LeftBracketingBar"                                          V                C                                            V                C                B                                      "RightBracketingBar"                                              (        1        )            
where Vc is the signal at the xe2x80x9cCxe2x80x9d port of the conversion circuit to be transmitted to the full-duplex port, e.g., a hybrid, (FIG. 1A), and VCB is the component of Vc appearing at the xe2x80x9cBxe2x80x9d port of the hybrid.
The quality of the impedance match between the balancing network and the local loop can also be measured by the return loss of the network. A low return loss indicates a poor impedance match or mismatch between the balancing network and the local loop. A network with a low return loss thus will suffer from more uncancelled near-end signal at the near end.
In most hybrid structures, good return loss between the balancing network and the local loop will result in good transhybrid loss. THL is the most important measure of a hybrid or echo canceller""s attenuation of near-end echo. Note that in some echo canceller realizations, impedances can be scaled so that the return loss between the balancing network and the cable will be small, but the THL will still be high.
Previous balancing networks for full-duplex telecommunication systems offered good transhybrid loss in the voice frequency band, but only fair transhybrid loss over the wider band of frequencies used for high bit-rate digital communications systems. Typical passive balance networks have been limited to two basic designs: simple R-C circuits and complex R-L-C circuits. Simple R-C designs have been limited in accuracy and in frequency response, while conventional R-L-C designs have proven to be more expensive, bulkier, and harder to integrate as compared to conventional R-C designs.
Because of the shortcomings of prior art cable balance networks, the need often arises for a simple and accurate representation of a cable""s input impedance. This is true especially in areas of telecommunication systems and echo canceller design and simulation. For example, cable balance networks utilizing complete multi-section resistance-inductance-conductance-capacitance (R-L-G-C) designs require too many discrete components and are extremely expensive to produce and manufacture because of the inductance requirement. Similarly, in the area of computer simulation, mathematical models of circuits based on the use of complex damped exponentials and R-L-G-C look-up tables require special programming and are computationally intensive and not readily adaptable to circuit simulators for use in computing the impedance of local loop cables.
Therefore, the need for simple and accurate balancing networks is especially critical when synthesizing or simulating communication system circuits using different wire gauges and cable lengths, or when tailoring an existing communication system for use in different frequency bands.
It is, therefore, a principal object of the present invention to provide a simple cable balancing network for a full-duplex telecommunication system that provides accurate cable input impedance modeling over a desired frequency range.
It is another object of the present invention to provide a simple, yet high performance cable balancing network for a full-duplex telecommunication system, where the improvement in transhybrid loss reduces the computational requirements and costs of an associated digital echo canceler.
It is yet another objective of the present invention to provide a method for optimizing the transhybrid loss provided for a full-duplex to half-duplex conversion circuit by a cable balance network.
It is still another object of the present invention to provide a method for synthesizing complex impedances and simulating useful telecommunications circuits by relatively simple circuit models thereof.
In accordance with the present invention, a cable balance network for echo cancelling in a full-duplex telecommunication system incorporating a full-duplex port, at least one half-duplex port, and a full-duplex to half-duplex conversion circuit, comprises: (a) a first plurality n of resistors Rn, each of said resistors Rn having a fixed resistance value dependent upon the impedance connected to the full-duplex port and each of said resistors Rn having a positive node and a negative node; (b) a second plurality nxe2x88x921 of capacitors Cnxe2x88x921, each of the capacitors Cnxe2x88x921 having a fixed capacitance value dependent upon the impedance connected to the full-duplex port and each of the capacitors Cnxe2x88x921 having a positive node and a negative node; and (c) a circuit configuration comprising at least two of the resistors and at least one of the capacitors such that all the negative nodes of each of the resistors are effectively connected together, and such that the capacitor is connected between the positive nodes of the at least two resistors for all values of n; the combination being so constructed and arranged that the circuit configuration, resistance values, and capacitance values define the cable balance network input impedance and an optimized transhybrid loss for the conversion circuit.
In further accordance with the present invention, a cable balance network for a full-duplex telecommunication system incorporating a full-duplex port, at least one half-duplex port, and a full-duplex to half-duplex conversion circuit, comprises: (a) four resistors R1, R2, R3, and R4, each of the resistors having a fixed resistance value dependent upon the impedance connected to the full-duplex port and each of the resistors having a positive node and a negative node; (b) three capacitors C1, C2, and C3, each of the capacitors having a fixed capacitance value dependent upon the impedance connected to the full-duplex port and each of the capacitors having a positive node and a negative node; and (c) a circuit configuration comprising the resistors and the capacitors, such that all of the negative nodes of the resistors are effectively connected together and such that the capacitor C1 is connected between the positive nodes of resistors R1 and R2, the capacitor C2 is connected between the positive nodes of resistors R2 and R3, and the capacitor C3 is connected between the positive nodes of resistors R3 and R4; the combination being so constructed and arranged that the circuit configuration, resistance values, and capacitance values define the cable balance network input impedance and an optimized transhybrid loss for the conversion circuit.
In further accordance with the present invention, a cable balance network for a full-duplex telecommunication system incorporating a full-duplex port, at least one half-duplex port, and a full-duplex to half-duplex conversion circuit, the cable balance network comprises: (a) five resistors R1, R2, R3, R4, and R5, each of the resistors having a fixed resistance value dependent upon the impedance connected to the full-duplex port and each of the resistors having a positive node and a negative node; (b) four capacitors C1,, C2, C3, and C4, each of the capacitors having a fixed capacitance value dependent upon the impedance connected to the full-duplex port and each of the capacitors having a positive node and a negative node; and (c) a circuit configuration comprising the resistors and the capacitors, such that all of the negative nodes of each of the resistors are effectively connected together and such that the capacitor C1 is connected between the positive nodes of resistors R1 and R2, the capacitor C2 is connected between the positive nodes of resistors R2 and R3, the capacitor C3 is connected between the positive nodes of resistors R3 and R4, and the capacitor C4 is connected between the positive nodes of resistors R4 and R5; the combination being so constructed and arranged that the circuit configuration, resistance values, and capacitance values define the cable balance network input impedance and an optimized transhybrid loss for the conversion circuit.
In further accordance with the present invention there is provided with respect to a cable balance network for a full-duplex telecommunication system incorporating a full duplex port, at least one half-duplex port, and a full-duplex to half-duplex conversion circuit, a method for synthesizing an R-C cable balance network by selecting resistance and capacitance values for each of the resistors and the capacitors in the balance network so as to optimize the input impedance match and transhybrid loss of the cable balance network, the method comprising the steps of: (a) characterizing the input impedance of the full-duplex port over a selected frequency range; (b) selecting as a resistance value of the network input resistor R1 the impedance connected to the full-duplex port at DC; (c) selecting as the resistance values of the remaining resistors in the network such that the resistance of the parallel combination of the resistors equals the impedance connected to the full-duplex port at high frequencies; and (d) optimizing the resistance and capacitance values to optimize the transhybrid loss for the conversion circuit.